Design and analysis of phase II cancer trials: A review of statistical methods and guidelines for medical researchers

L. Mariani, E. Marubini

Research output: Contribution to journalArticle

27 Citations (Scopus)

Abstract

A number of statistical methods have been proposed in the last three decades for the design and analysis of phase II trials in the oncological field. Herein, a review of these methods is supplied. Practical guidelines for the proper choice of study design and some worked examples are also given. Based on this material, it appears that one-sample studies are likely to yield reliable results in early phase II testing or when the trial is of a comparative nature, provided that an accurate estimate of the response rate is available for standard treatment. Multi-stage designs are useful for reducing, on the average, the number of patients required. In situations different from those cited herein, controlled trials are preferable but require a larger number of patients. This drawback can be lessened through inclusion in the analysis of information from historical controls or the adoption of early stopping rules.

Original languageEnglish
Pages (from-to)61-88
Number of pages28
JournalInternational Statistical Review
Volume64
Issue number1
Publication statusPublished - Apr 1996

Fingerprint

Statistical method
Cancer
Early Stopping
Stopping Rule
Inclusion
Likely
Testing
Estimate
Review
Design
Statistical methods
Standards
Response rate
Stopping rule

Keywords

  • Binary outcomes
  • Cancer chemotherapy
  • Clinical research
  • Phase II trials
  • Sample size

ASJC Scopus subject areas

  • Mathematics(all)
  • Statistics and Probability

Cite this

Design and analysis of phase II cancer trials : A review of statistical methods and guidelines for medical researchers. / Mariani, L.; Marubini, E.

In: International Statistical Review, Vol. 64, No. 1, 04.1996, p. 61-88.

Research output: Contribution to journalArticle

@article{45361f568cb44bb78bb9e645c7379b6e,
title = "Design and analysis of phase II cancer trials: A review of statistical methods and guidelines for medical researchers",
abstract = "A number of statistical methods have been proposed in the last three decades for the design and analysis of phase II trials in the oncological field. Herein, a review of these methods is supplied. Practical guidelines for the proper choice of study design and some worked examples are also given. Based on this material, it appears that one-sample studies are likely to yield reliable results in early phase II testing or when the trial is of a comparative nature, provided that an accurate estimate of the response rate is available for standard treatment. Multi-stage designs are useful for reducing, on the average, the number of patients required. In situations different from those cited herein, controlled trials are preferable but require a larger number of patients. This drawback can be lessened through inclusion in the analysis of information from historical controls or the adoption of early stopping rules.",
keywords = "Binary outcomes, Cancer chemotherapy, Clinical research, Phase II trials, Sample size",
author = "L. Mariani and E. Marubini",
year = "1996",
month = "4",
language = "English",
volume = "64",
pages = "61--88",
journal = "International Statistical Review",
issn = "0306-7734",
publisher = "International Statistical Institute",
number = "1",

}

TY - JOUR

T1 - Design and analysis of phase II cancer trials

T2 - A review of statistical methods and guidelines for medical researchers

AU - Mariani, L.

AU - Marubini, E.

PY - 1996/4

Y1 - 1996/4

N2 - A number of statistical methods have been proposed in the last three decades for the design and analysis of phase II trials in the oncological field. Herein, a review of these methods is supplied. Practical guidelines for the proper choice of study design and some worked examples are also given. Based on this material, it appears that one-sample studies are likely to yield reliable results in early phase II testing or when the trial is of a comparative nature, provided that an accurate estimate of the response rate is available for standard treatment. Multi-stage designs are useful for reducing, on the average, the number of patients required. In situations different from those cited herein, controlled trials are preferable but require a larger number of patients. This drawback can be lessened through inclusion in the analysis of information from historical controls or the adoption of early stopping rules.

AB - A number of statistical methods have been proposed in the last three decades for the design and analysis of phase II trials in the oncological field. Herein, a review of these methods is supplied. Practical guidelines for the proper choice of study design and some worked examples are also given. Based on this material, it appears that one-sample studies are likely to yield reliable results in early phase II testing or when the trial is of a comparative nature, provided that an accurate estimate of the response rate is available for standard treatment. Multi-stage designs are useful for reducing, on the average, the number of patients required. In situations different from those cited herein, controlled trials are preferable but require a larger number of patients. This drawback can be lessened through inclusion in the analysis of information from historical controls or the adoption of early stopping rules.

KW - Binary outcomes

KW - Cancer chemotherapy

KW - Clinical research

KW - Phase II trials

KW - Sample size

UR - http://www.scopus.com/inward/record.url?scp=0030557614&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030557614&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0030557614

VL - 64

SP - 61

EP - 88

JO - International Statistical Review

JF - International Statistical Review

SN - 0306-7734

IS - 1

ER -